Introduction
For the last six months, Ive avoided the usual payment infrastructure headaches by building stores that rely on NOWPayments for crypto and mobile money, and Supabase as the single source of truth for orders. You dont need a US bank account. You dont need to beg a payment processor for access. You only need an idea, a domain, and the will to ship.
This setup has helped me test pricing, run limited drops, and even run a store that sold digital templates to 234 customers across Nigeria, Ghana, and Kenya in under two weeks. Below Ill share the exact stack, the wiring logic, and the deployment steps that actually worked in places where infrastructure fights back.
Why Supabase First
Ive tried MongoDB Atlas, Firebase Realtime, and even a self-hosted PostgreSQL on a $5 DO droplet. All of them required extra glue code for realtime updates, manual backups, or firewall rules to keep bots out. Supabase bundles Postgres, realtime subscriptions, row-level security, and a free tier that scales to 50,000 rows. That threshold alone covered my first six months of sales data without ever hitting the limit.
Heres what matters:
- You get a Postgres table called orders automatically. - Row-level security lets you lock every row to only the admin or the customer who placed it. - The realtime API pushes new orders to your dashboard without polling. - The Storage bucket keeps receipts and artwork—no external CDN fees.
I started with a single orders table:
id uuid primary key default gen_random_uuid(), email text not null, product_name text not null, tx_ref text unique, payment_status text default pending, amount numeric(10,2), currency text default NGN, crypto_address text, created_at timestamp with time zone default now()
Add indexes on tx_ref and created_at. That single table became my orderbook, my analytics source, and my customer list merged into one.
Frontend Stack That Doesnt Hate Africa
I run the store on GitHub Pages for zero hosting costs and instant global CDN. The static site is built with 11ty, so no Node server fees. All API calls go to Supabase Edge Functions, which live on Deno under Cloudflares edge network.
Steps:
- Push code to GitHub, enable GitHub Pages (set source to /docs). - Create a new Supabase project, go to SQL Editor, paste the table above. - Create an Edge Function called create-order that accepts POST {email, product_slug, amount, currency}. - In the Edge Function, insert a row into orders, generate a NOWPayments invoice link, and return the link to the frontend. - Use Cloudflare Workers KV to cache product slugs and prices so page loads stay under 200 ms in Lagos.
No Node.js hosting, no Docker, no Nginx config to maintain. The whole system stays inside GitHub, Supabase, and Cloudflare—three services that actually respond to African traffic.
Payments Without Stripe or PayPal
NOWPayments covers Nigeria (Naira), Kenya (KES), and Ghana (GHS) with mobile money and crypto rails. I send the amount and currency from the frontend, NOWPayments returns an invoice ID and a payment link. I store the invoice ID in tx_ref so Supabase later knows which NOWPayments webhook belongs to which order.
Inside Supabase Edge Functions, I listen to NOWPayments webhooks:
When paid status arrives, the function updates payment_status to completed and triggers a realtime event that pushes the order to my Slack channel via a Cloudflare Worker that forwards the payload.
No payment gateway dashboard needed; everything is in code and Slack alerts.
Security and Backups You Can Sleep On
Supabase gives row-level security policies out of the box. I created a policy that only lets customers see their own rows:
create policy customer_can_read ON orders for select using (auth.uid() = customer_id);
I also turned on automatic backups every 24 hours, sending PostgreSQL dumps to a cheap Wasabi bucket. That backup script is a single Cloudflare Worker scheduled with cron triggers—no extra cron job on a server.
I sleep because the entire stack is declarative: schema in a SQL file, policies in SQL, backup script in Workers. If my laptop dies, I can recreate the entire store from GitHub and Supabase in under an hour.
Launch Workflow: 48 Hours or Less
Day 0 Pick three products and price them in NGN. Create the products table in Supabase and seed it.
Day 1 Build the 11ty templates, wire up the Edge Function, and test NOWPayments sandbox with 200 Naira orders.
Day 2 Buy a domain on Namecheap, set up Cloudflare DNS with proxy on, push to GitHub Pages, switch sandbox to live keys. Send the first email campaign to my small list. First sale usually arrives within four hours.
Day 3 Wake up, see 15 orders in Supabase realtime table, run a quick analytics query in the SQL editor, export CSV to Google Sheets, and call it done.
This pattern has worked in multiple markets. The only variables are product-market fit and the speed of marketing—not infrastructure luck.
Conclusion
I built this stack because